{"id":44545,"date":"2024-08-01T07:02:14","date_gmt":"2024-08-01T07:02:14","guid":{"rendered":"https:\/\/meetcody.ai\/blog\/architecture-gemma-2-2b-innovations-et-applications\/"},"modified":"2024-08-01T07:02:14","modified_gmt":"2024-08-01T07:02:14","slug":"architecture-gemma-2-2b-innovations-et-applications","status":"publish","type":"post","link":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/","title":{"rendered":"Architecture Gemma 2 2b : Innovations et applications"},"content":{"rendered":"
\n

R\u00e9cemment, Google DeepMind a d\u00e9voil\u00e9 le dernier n\u00e9 de son impressionnante gamme de mod\u00e8les d’IA : Gemma 2 2b.
\nCe mod\u00e8le, avec ses 2 milliards de param\u00e8tres, marque une \u00e9tape importante dans le d\u00e9veloppement de solutions d’IA avanc\u00e9es mais compactes.
\nGemma 2 2b a \u00e9t\u00e9 con\u00e7u pour offrir les meilleures performances de sa cat\u00e9gorie malgr\u00e9 sa taille relativement petite par rapport aux autres mod\u00e8les de l’industrie. <\/span><\/p>\n<\/div>\n

Ce qui distingue Gemma 2 2b, c’est sa capacit\u00e9 exceptionnelle \u00e0 surpasser ses homologues nettement plus grands.
\nIl a notamment surpass\u00e9 le GPT-3.5 et le Mixtral 8x7B dans divers tests de r\u00e9f\u00e9rence, d\u00e9montrant ainsi son efficacit\u00e9 et sa robustesse sup\u00e9rieures.
\nIl s’agit d’une nouvelle norme de performance en mati\u00e8re d’IA, qui prouve que plus c’est gros, plus c’est bon. <\/p>\n

\n

Caract\u00e9ristiques principales de Gemma 2b\"Les<\/h2>\n<\/div>\n

Gemma 2 2b se distingue par ses performances impressionnantes.
\nAvec un score de 1130 dans le LMSYS Chatbot Arena, il surpasse des mod\u00e8les beaucoup plus grands comme GPT-3.5-Turbo-0613 (1117) et Mixtral-8x7B (1114), d\u00e9montrant l’efficacit\u00e9 des techniques d’apprentissage sophistiqu\u00e9es par rapport \u00e0 la simple taille des param\u00e8tres.
\nCes rep\u00e8res montrent que Gemma 2 2b n’est pas seulement puissant, mais aussi tr\u00e8s efficace.
\nLes classificateurs de s\u00e9curit\u00e9 ShieldGemma, con\u00e7us pour d\u00e9tecter et mod\u00e9rer les contenus nuisibles, constituent une autre caract\u00e9ristique essentielle.
\nShieldGemma cible diverses cat\u00e9gories telles que les discours haineux et les contenus sexuellement explicites, garantissant ainsi des interactions plus s\u00fbres avec l’IA.
\nGemma 2 2b se positionne ainsi comme une solution d’IA fiable pour les applications sensibles.
\nEn outre, Gemma Scope<\/a> am\u00e9liore la transparence en utilisant des autoencodeurs \u00e9pars pour d\u00e9composer les processus de prise de d\u00e9cision du mod\u00e8le.
\nCet outil permet de comprendre clairement comment l’architecture de Gemma 2 2b traite l’information, contribuant ainsi au d\u00e9veloppement de syst\u00e8mes d’IA plus transparents et plus fiables. <\/p>\n

\n

Gemma 2b Architecture<\/h2>\n<\/div>\n

L’architecture Gemma 2 2b s’appuie sur des techniques avanc\u00e9es de compression et de distillation de mod\u00e8les pour obtenir des performances sup\u00e9rieures malgr\u00e9 sa taille compacte.
\nCes m\u00e9thodes permettent au mod\u00e8le de distiller les connaissances de ses pr\u00e9d\u00e9cesseurs de plus grande taille, ce qui donne un syst\u00e8me d’IA tr\u00e8s efficace et puissant.
\nGemma 2 2b a \u00e9t\u00e9 entra\u00een\u00e9 sur un important ensemble de donn\u00e9es comprenant 2 billions de jetons, en utilisant le mat\u00e9riel TPU v5e de Google \u00e0 la pointe de la technologie.
\nCela permet un entra\u00eenement rapide et efficace, garantissant que le mod\u00e8le peut g\u00e9rer des t\u00e2ches diverses et complexes dans plusieurs langues.
\nCompar\u00e9 aux autres mod\u00e8les de la famille Gemma, tels que les variantes \u00e0 9 milliards (9B) et 27 milliards (27B) de param\u00e8tres, Gemma 2 2b se distingue par son \u00e9quilibre entre taille et efficacit\u00e9.
\nSon architecture est con\u00e7ue pour fonctionner de mani\u00e8re exceptionnelle sur une large gamme de mat\u00e9riel, des ordinateurs portables aux d\u00e9ploiements en nuage, ce qui en fait un choix polyvalent pour les chercheurs et les d\u00e9veloppeurs. <\/p>\n

\n

Gemma 2 2b change-t-il vraiment la donne ?<\/h2>\n<\/div>\n

L’architecture Gemma 2 2b a des implications significatives pour l’IA mobile et l’informatique de pointe.
\nSa taille compacte lui permet d’\u00eatre d\u00e9ploy\u00e9e sur divers appareils grand public sans sacrifier les performances, ce qui ouvre de nouvelles possibilit\u00e9s dans les smartphones et autres gadgets portables.
\nUn autre aspect essentiel de Gemma 2 2b est son int\u00e9r\u00eat pour l’environnement.
\nDes mod\u00e8les plus petits et plus efficaces comme Gemma 2 2b r\u00e9duisent la puissance de calcul<\/a> n\u00e9cessaire \u00e0 la formation et au d\u00e9ploiement, ce qui r\u00e9pond aux pr\u00e9occupations croissantes concernant l’empreinte carbone des grands syst\u00e8mes d’IA.
\nCette efficacit\u00e9 ne rend pas seulement l’IA plus accessible, elle soutient \u00e9galement le d\u00e9veloppement technologique durable.
\n\u00c0 l’avenir, l’innovation continue sera essentielle pour que Google conserve son avantage concurrentiel.
\nAlors que d’autres g\u00e9ants de la technologie sortent des mod\u00e8les plus r\u00e9cents et plus avanc\u00e9s, comme le Llama 3.1 de Meta et le GPT-4o<\/a> d’OpenAI, Google doit se concentrer sur le perfectionnement de la s\u00e9rie Gemma.
\nLes am\u00e9liorations potentielles pourraient inclure des fonctions de s\u00e9curit\u00e9 am\u00e9lior\u00e9es comme ShieldGemma et une plus grande transparence gr\u00e2ce \u00e0 des outils comme Gemma Scope. <\/p>\n

\n

Conclusion : La voie \u00e0 suivre pour Gemma 2b<\/h2>\n<\/div>\n

L’architecture Gemma 2 2b a \u00e9tabli une nouvelle r\u00e9f\u00e9rence dans le paysage de l’intelligence artificielle gr\u00e2ce \u00e0 ses performances impressionnantes dans diverses configurations mat\u00e9rielles, tout en conservant un nombre compact de param\u00e8tres.
\nSurpassant des mod\u00e8les plus importants comme GPT-3.5 et Mixtral 8x7b, Gemma 2 2b prouve que des architectures efficaces et des ensembles de donn\u00e9es de haute qualit\u00e9 peuvent rivaliser avec la taille brute des param\u00e8tres.
\nL’une des principales r\u00e9alisations de Gemma 2 2b est la d\u00e9mocratisation de la technologie de l’IA.
\nEn rendant le mod\u00e8le disponible sur des appareils grand public et des plateformes open-source telles que Hugging Face, Google soutient un \u00e9cosyst\u00e8me d’IA plus accessible et plus transparent.
\nDes outils tels que ShieldGemma et Gemma Scope contribuent \u00e0 rendre les applications d’IA plus s\u00fbres<\/a> et plus transparentes.
\nAu fur et \u00e0 mesure que l’IA \u00e9volue, l’accent pourrait \u00eatre mis non plus sur la cr\u00e9ation de mod\u00e8les toujours plus grands, mais sur le perfectionnement de mod\u00e8les plus petits et plus efficaces.
\nGemma 2 2b marque un tournant dans cette direction, en favorisant la durabilit\u00e9 et l’accessibilit\u00e9.
\nCela pourrait annoncer une nouvelle \u00e8re dans laquelle des mod\u00e8les d’IA tr\u00e8s performants pourraient fonctionner efficacement sur une grande vari\u00e9t\u00e9 de mat\u00e9riel, d\u00e9mocratisant ainsi les capacit\u00e9s d’IA avanc\u00e9es.
\nEn r\u00e9sum\u00e9, le succ\u00e8s de Gemma 2 2b ouvre une voie prometteuse pour l’avenir de l’IA.
\nAlors que Google continue d’innover, les progr\u00e8s r\u00e9alis\u00e9s dans le domaine des mod\u00e8les d’IA efficaces et accessibles devraient faire progresser le secteur, en \u00e9largissant les horizons de ce que l’IA peut accomplir au niveau mondial. <\/p>\n","protected":false},"excerpt":{"rendered":"

R\u00e9cemment, Google DeepMind a d\u00e9voil\u00e9 le dernier n\u00e9 de son impressionnante gamme de mod\u00e8les d’IA : Gemma 2 2b. Ce mod\u00e8le, avec ses 2 milliards de param\u00e8tres, marque une \u00e9tape importante dans le d\u00e9veloppement de solutions d’IA avanc\u00e9es mais compactes. Gemma 2 2b a \u00e9t\u00e9 con\u00e7u pour offrir les meilleures performances de sa cat\u00e9gorie malgr\u00e9… Read more »<\/a><\/p>\n","protected":false},"author":2,"featured_media":44537,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"footnotes":""},"categories":[29],"tags":[],"class_list":["post-44545","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-non-classifiee"],"acf":[],"yoast_head":"\nArchitecture Gemma 2 2b : Innovations et applications<\/title>\n<meta name=\"description\" content=\"D\u00e9couvrez Gemma 2 2b : Le mod\u00e8le d'IA compact mais puissant de Google DeepMind qui surpasse ses homologues plus grands dans les tests de r\u00e9f\u00e9rence.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Architecture Gemma 2 2b : Innovations et applications\" \/>\n<meta property=\"og:description\" content=\"D\u00e9couvrez Gemma 2 2b : Le mod\u00e8le d'IA compact mais puissant de Google DeepMind qui surpasse ses homologues plus grands dans les tests de r\u00e9f\u00e9rence.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\" \/>\n<meta property=\"og:site_name\" content=\"Cody - The AI Trained on Your Business\" \/>\n<meta property=\"article:published_time\" content=\"2024-08-01T07:02:14+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png\" \/>\n\t<meta property=\"og:image:width\" content=\"2880\" \/>\n\t<meta property=\"og:image:height\" content=\"1620\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Om Kamath\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@meetcodyai\" \/>\n<meta name=\"twitter:site\" content=\"@meetcodyai\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Om Kamath\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"5 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\"},\"author\":{\"name\":\"Om Kamath\",\"@id\":\"https:\/\/meetcody.ai\/#\/schema\/person\/cde65ec55b79cd833a9777d0a62e83c8\"},\"headline\":\"Architecture Gemma 2 2b : Innovations et applications\",\"datePublished\":\"2024-08-01T07:02:14+00:00\",\"dateModified\":\"2024-08-01T07:02:14+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\"},\"wordCount\":1054,\"publisher\":{\"@id\":\"https:\/\/meetcody.ai\/#organization\"},\"image\":{\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png\",\"articleSection\":[\"Non classifi\u00e9(e)\"],\"inLanguage\":\"fr-FR\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\",\"url\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\",\"name\":\"Architecture Gemma 2 2b : Innovations et applications\",\"isPartOf\":{\"@id\":\"https:\/\/meetcody.ai\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png\",\"datePublished\":\"2024-08-01T07:02:14+00:00\",\"dateModified\":\"2024-08-01T07:02:14+00:00\",\"description\":\"D\u00e9couvrez Gemma 2 2b : Le mod\u00e8le d'IA compact mais puissant de Google DeepMind qui surpasse ses homologues plus grands dans les tests de r\u00e9f\u00e9rence.\",\"breadcrumb\":{\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage\",\"url\":\"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png\",\"contentUrl\":\"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png\",\"width\":2880,\"height\":1620},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/meetcody.ai\/fr\/home-v2\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Architecture Gemma 2 2b : Innovations et applications\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/meetcody.ai\/#website\",\"url\":\"https:\/\/meetcody.ai\/\",\"name\":\"Cody AI - The AI Trained on Your Business\",\"description\":\"AI Powered Knowledge Base for Employees\",\"publisher\":{\"@id\":\"https:\/\/meetcody.ai\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/meetcody.ai\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/meetcody.ai\/#organization\",\"name\":\"Cody AI - The AI Trained on Your Business\",\"url\":\"https:\/\/meetcody.ai\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/meetcody.ai\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/meetcody.ai\/wp-content\/uploads\/2023\/05\/logo-codyai.svg\",\"contentUrl\":\"https:\/\/meetcody.ai\/wp-content\/uploads\/2023\/05\/logo-codyai.svg\",\"width\":\"1024\",\"height\":\"1024\",\"caption\":\"Cody AI - The AI Trained on Your Business\"},\"image\":{\"@id\":\"https:\/\/meetcody.ai\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/x.com\/meetcodyai\",\"https:\/\/discord.com\/invite\/jXEVDcFxqs\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/meetcody.ai\/#\/schema\/person\/cde65ec55b79cd833a9777d0a62e83c8\",\"name\":\"Om Kamath\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/meetcody.ai\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/3903c678cd7f6c8df0a843ae177998f5d413954afa3062f984a030a889a97849?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/3903c678cd7f6c8df0a843ae177998f5d413954afa3062f984a030a889a97849?s=96&d=mm&r=g\",\"caption\":\"Om Kamath\"},\"description\":\"Om Kamath\",\"sameAs\":[\"http:\/\/meetcody.ai\"],\"url\":\"https:\/\/meetcody.ai\/fr\/blog\/author\/omkamath\/\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Architecture Gemma 2 2b : Innovations et applications","description":"D\u00e9couvrez Gemma 2 2b : Le mod\u00e8le d'IA compact mais puissant de Google DeepMind qui surpasse ses homologues plus grands dans les tests de r\u00e9f\u00e9rence.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/","og_locale":"fr_FR","og_type":"article","og_title":"Architecture Gemma 2 2b : Innovations et applications","og_description":"D\u00e9couvrez Gemma 2 2b : Le mod\u00e8le d'IA compact mais puissant de Google DeepMind qui surpasse ses homologues plus grands dans les tests de r\u00e9f\u00e9rence.","og_url":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/","og_site_name":"Cody - The AI Trained on Your Business","article_published_time":"2024-08-01T07:02:14+00:00","og_image":[{"width":2880,"height":1620,"url":"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png","type":"image\/png"}],"author":"Om Kamath","twitter_card":"summary_large_image","twitter_creator":"@meetcodyai","twitter_site":"@meetcodyai","twitter_misc":{"Written by":"Om Kamath","Est. reading time":"5 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#article","isPartOf":{"@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/"},"author":{"name":"Om Kamath","@id":"https:\/\/meetcody.ai\/#\/schema\/person\/cde65ec55b79cd833a9777d0a62e83c8"},"headline":"Architecture Gemma 2 2b : Innovations et applications","datePublished":"2024-08-01T07:02:14+00:00","dateModified":"2024-08-01T07:02:14+00:00","mainEntityOfPage":{"@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/"},"wordCount":1054,"publisher":{"@id":"https:\/\/meetcody.ai\/#organization"},"image":{"@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage"},"thumbnailUrl":"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png","articleSection":["Non classifi\u00e9(e)"],"inLanguage":"fr-FR"},{"@type":"WebPage","@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/","url":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/","name":"Architecture Gemma 2 2b : Innovations et applications","isPartOf":{"@id":"https:\/\/meetcody.ai\/#website"},"primaryImageOfPage":{"@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage"},"image":{"@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage"},"thumbnailUrl":"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png","datePublished":"2024-08-01T07:02:14+00:00","dateModified":"2024-08-01T07:02:14+00:00","description":"D\u00e9couvrez Gemma 2 2b : Le mod\u00e8le d'IA compact mais puissant de Google DeepMind qui surpasse ses homologues plus grands dans les tests de r\u00e9f\u00e9rence.","breadcrumb":{"@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/"]}]},{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#primaryimage","url":"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png","contentUrl":"https:\/\/meetcody.ai\/wp-content\/uploads\/2024\/08\/Gemma.png","width":2880,"height":1620},{"@type":"BreadcrumbList","@id":"https:\/\/meetcody.ai\/fr\/blog\/architecture-gemma-2-2b-innovations-et-applications\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/meetcody.ai\/fr\/home-v2\/"},{"@type":"ListItem","position":2,"name":"Architecture Gemma 2 2b : Innovations et applications"}]},{"@type":"WebSite","@id":"https:\/\/meetcody.ai\/#website","url":"https:\/\/meetcody.ai\/","name":"Cody AI - The AI Trained on Your Business","description":"AI Powered Knowledge Base for Employees","publisher":{"@id":"https:\/\/meetcody.ai\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/meetcody.ai\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"fr-FR"},{"@type":"Organization","@id":"https:\/\/meetcody.ai\/#organization","name":"Cody AI - The AI Trained on Your Business","url":"https:\/\/meetcody.ai\/","logo":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/meetcody.ai\/#\/schema\/logo\/image\/","url":"https:\/\/meetcody.ai\/wp-content\/uploads\/2023\/05\/logo-codyai.svg","contentUrl":"https:\/\/meetcody.ai\/wp-content\/uploads\/2023\/05\/logo-codyai.svg","width":"1024","height":"1024","caption":"Cody AI - The AI Trained on Your Business"},"image":{"@id":"https:\/\/meetcody.ai\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/meetcodyai","https:\/\/discord.com\/invite\/jXEVDcFxqs"]},{"@type":"Person","@id":"https:\/\/meetcody.ai\/#\/schema\/person\/cde65ec55b79cd833a9777d0a62e83c8","name":"Om Kamath","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/meetcody.ai\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/3903c678cd7f6c8df0a843ae177998f5d413954afa3062f984a030a889a97849?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/3903c678cd7f6c8df0a843ae177998f5d413954afa3062f984a030a889a97849?s=96&d=mm&r=g","caption":"Om Kamath"},"description":"Om Kamath","sameAs":["http:\/\/meetcody.ai"],"url":"https:\/\/meetcody.ai\/fr\/blog\/author\/omkamath\/"}]}},"_links":{"self":[{"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/posts\/44545","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/comments?post=44545"}],"version-history":[{"count":0,"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/posts\/44545\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/media\/44537"}],"wp:attachment":[{"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/media?parent=44545"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/categories?post=44545"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/meetcody.ai\/fr\/wp-json\/wp\/v2\/tags?post=44545"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}